项目名称: 几何与图像计算中的变分方法与算法
项目编号: No.11301289
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 数理科学和化学
项目作者: 吴春林
作者单位: 南开大学
项目金额: 22万元
中文摘要: 近年来,变分方法在几何与图像计算中发挥了越来越重要的作用,被成功应用于各种几何计算与图像处理问题中。在本项目中,我们将利用变分方法,同时开展网格曲面处理以及平面图像处理方面的研究工作。我们将通过网格曲面上的分片常值函数空间,设计各种变分模型与快速算法,并运用到网格曲面的去噪处理问题;我们将考虑网格曲面的基于面(即网格面元)而不是网格顶点的分割问题,针对surface-type分割与part-type分割分别设计变分模型与快速算法;我们将系统地研究透射式成像时的图像去噪问题,包括变分模型的建立与有效数值算法的设计。这些问题是计算几何及图像处理中的重要问题。它们的解决将为几何建模与图像处理提供更有效的算法与更广阔的工业应用前景。项目申请人在几何与图像计算中的变分方法与算法方面有很好的基础,有望在所提出的问题上取得实质性的进展。
中文关键词: 网格曲面处理;图像处理;分片常数函数空间;变分模型;优化算法
英文摘要: In recent years, variational methods have been playing more and more important roles in geometry and image computing. They have been successfully applied to various geometry computing and image processing problems. In this project, we will use variational methods to investigate on both mesh surface processing and planar image processing. Starting from piecewise constant function spaces on mesh surfaces, we will design various variational models and efficient algorithms, and then apply these models and algorithms to mesh surface denoising. Besides, we will consider mesh segmentation problem where a given mesh is to be partitioned as several parts containing faces instead of mesh vertices. Variational segmentation models and efficient algorithms will be designed for both surface-type segmentation and part-type segmentation. In addition, we will construct variational models and efficient algorithms for denoising images generated by transmission imaging. The problems involved in this project are of great importance in geometry computing and image processing, which shall contribute more efficient algorithms for geometry modelling and image processing and thus open a broader space in industry applications. The applicants have solid background in variational methods and algorithms for geometry and image computing and a
英文关键词: mesh surface processing;image processing;piecewise constant function space;variational model;optimization algorithm